3. Results.
In table 1 we show the main characteristics of the critical COVID-19 patients aged ≤60 years or older. We found a higher frequency of male in the two age groups. As expected the frequency of hypertensives, diabetics, and hyperlipaemics was higher among the patients >60 years. There were a total of 103 deaths (table 1 ). The frequency of mtDNA 7028C that characterises haplogroup H was significantly lower in the ≤60 years patients (38% vs 48%). The frequency of 16223T was significantly higher in the ≤60 years group (18% vs 10%). The 7028C and 16223T did not differ between the two age control groups (table 2 ). Thus, we concluded that the difference between pagtients aged ≤60 and >60 years would not be a consequence of a survival-effect in the general population, and these mtDNA markers could thus represent risk factors for age-dependent critical COVID-19.
Control region length heteroplasmy. Fluorescent capillary electrophoresis of the amplified fragments showed multiple peaks (heteroplasmy) at the 16184-16193 poly-C tract significantly more frequent among the ≤60 years patients compared to age matched controls (19% vs 9%; p=0.002). Heteroplasmy was also more common among the ≤60 years compared to the older patients (19% vs 10%; p=0.02), without differences between patients and controls >60 years (table 2 ). We found a maximum frequency of length heteroplasmy among patients aged <50 years, suggesting that this mitochondrial genomic event was significantly associated with critical COVID-19 at younger age (Figure 2; suppl. Table 1 ).
The 16184-16193 poly-C tract that increases length heteroplasmy has been associated with several multifactorial disorders, including metabolic and cardiovascular. Because critical COVID-19 was significantly associated with hypertension, diabetes, dyslipidaemia, and male sex, we determined whether length heteroplasmy was higher among patients with these conditions. We did not find statistical differences between the groups (suppl. table 2 ). We also compared death and survivors, without significant difference. Moreover, after correcting by multiple variables advanced age was the only significantly associated with the risk of death (p<0.001).
We also determined the frequencies of the control region CAn-repeat. We did not find significant differences between patients and controls in the two age-groups. Heteroplasmy at this region was found in 3-6% of the study cohorts, without significant difference between the groups.
Poly-C heteroplasmy was linked to 16223 T. We found a higher frequency of poly-C heteroplasmy among patients 16223 T compared to 16223 C, in the two age groups (suppl. table 3 ). The 16223 T is the ancestral African allele and is present in less than 10% of current Europeans with the rare IXW haplogroups (suppl. table 4 ). This pointed to the possibility that the higher frequency of 16189 T and 162184-16193 poly-C heteroplasmy were a consequence of the linkage between 16223 T and 16189 C, a variant that increases the risk of poly-C instability (suppl. table 5 ).